Skip to content

Workstream 3 RFC: Sector-specific CoSAI Secure AI Risk Assessments #4

@davidlabianca

Description

@davidlabianca

Sector-specific CoSAI Secure AI Risk Assessments

Authors:

Summary

Identify and establish collaborations with sector-specific expert organizations to develop the first two sector-specific implementations of the CoSAI Secure AI Risk Assessment.

Through working with sector-specific expert groups CoSAI will be able to expand the core Risk Assessment to provide the appropriate flexibility required to implement sector-specific implementations as well as collaboratively build two sector-specific implementations with sector experts.

This effort will provide the needed learnings and improvements to allow broader adoption of the Risk Assessment through sector-specific tailoring efforts.

Priority

P1: This is important to include in the next release from this workstream.

Level of Effort

Medium: This will take a week or two to document.

Drawbacks

These efforts, by definition, requires outreach and relationship building with other communities of experts in order to create the virtual team required to deliver a sector-specific Secure AI Risk Assessment. Due to the complexity of x-organization efforts we may experience significant delays in pursuing these engagements which could delay the evolution and expansion of our Risk Map and Assessment.

Note: While this proposed effort is semi-dependent upon the proposed Workstream 3 RFC: CoSAI Secure AI Risk Map and Risk Assessment parallel execution and development should be possible given the source materials, for the artifacts, are pre-existing and the improvements from the upstream effort can be merged as they are developed.

Alternatives

There are many AI risk taxonomies that address the wide scope of issues and challenges within the AI domain. The breadth of many of these approaches increases the effort required by sector-specific security organizations and technical teams to identify and address the most critical security issues related to AI systems.

Reference Material & Prior Art

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type
    No fields configured for issues without a type.

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions